A protocol for fluorescent, flow cytometric quantification of senescent cancer cells induced by chemotherapy drugs in cell culture or in murine tumor models is presented. Optional procedures include co-immunostaining, sample fixation to facilitate large batch or time point analysis, and the enrichment of viable senescent cells by flow cytometric sorting.
Cellular senescence is a state of proliferative arrest induced by biological damage that normally accrues over years in aging cells but may also emerge rapidly in tumor cells as a response to damage induced by various cancer treatments. Tumor cell senescence is generally considered undesirable, as senescent cells become resistant to death and block tumor remission while exacerbating tumor malignancy and treatment resistance. Therefore, the identification of senescent tumor cells is of ongoing interest to the cancer research community. Various senescence assays exist, many based on the activity of the well-known senescence marker, senescence-associated beta-galactosidase (SA-β-Gal).
Typically, the SA-β-Gal assay is performed using a chromogenic substrate (X-Gal) on fixed cells, with the slow and subjective enumeration of “blue” senescent cells by light microscopy. Improved assays using cell-permeant, fluorescent SA-β-Gal substrates, including C12-FDG (green) and DDAO-Galactoside (DDAOG; far-red), have enabled the analysis of living cells and allowed the use of high-throughput fluorescent analysis platforms, including flow cytometers. C12-FDG is a well-documented probe for SA-β-Gal, but its green fluorescent emission overlaps with intrinsic cellular autofluorescence (AF) that arises during senescence due to the accumulation of lipofuscin aggregates. By utilizing the far-red SA-β-Gal probe DDAOG, green cellular autofluorescence can be used as a secondary parameter to confirm senescence, adding reliability to the assay. The remaining fluorescence channels can be used for cell viability staining or optional fluorescent immunolabeling.
Using flow cytometry, we demonstrate the use of DDAOG and lipofuscin autofluorescence as a dual-parameter assay for the identification of senescent tumor cells. Quantitation of the percentage of viable senescent cells is performed. If desired, an optional immunolabeling step may be included to evaluate cell surface antigens of interest. Identified senescent cells can also be flow cytometrically sorted and collected for downstream analysis. Collected senescent cells can be immediately lysed (e.g., for immunoassays or ‘omics analysis) or further cultured.
Senescent cells normally accumulate in organisms over years during normal biological aging but may also develop rapidly in tumor cells as a response to damage induced by various cancer treatments, including radiation and chemotherapy. Though no longer proliferating, therapy-induced senescent (TIS) tumor cells may contribute to treatment resistance and drive recurrence1,2,3. Factors secreted by TIS cells can exacerbate tumor malignancy by promoting immune evasion or metastasis4,5. TIS cells develop complex, context-specific phenotypes, altered metabolic profiles, and unique immune responses6,7,8. Therefore, the identification and characterization of TIS tumor cells induced by various cancer treatment approaches is a topic of ongoing interest to the cancer research community.
To detect TIS tumor cells, conventional senescence assays are widely used, primarily based on detecting increased activity of the senescence marker enzyme, the lysosomal beta-galactosidase GLB19. Detection at a near-neutral (rather than acidic) lysosomal pH allows for specific detection of senescence-associated beta-galactosidase (SA-β-Gal)10. A standard SA-β-Gal assay that has been used for several decades uses X-Gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside), a blue chromogenic beta-galactosidase substrate, to detect SA-β-Gal in fixed cells by light microscopy11. The X-Gal assay allows the qualitative visual confirmation of TIS utilizing commonly available reagents and laboratory equipment. A basic transmitted light microscope is the only instrumentation required to evaluate the presence of the blue chromogen. However, the X-Gal staining procedure can lack sensitivity, sometimes requiring more than 24 h for color to develop. Staining is followed by low-throughput, subjective scoring of individual senescent cells based on counting the cells exhibiting some level of intensity of the blue chromogen under a light microscope. As X-Gal is cell-impermeable, this assay requires solvent-fixed cells, which cannot be recovered for downstream analysis. When working with limited samples from animals or patients, this can be a major drawback.
Improved SA-β-Gal assays using cell-permeant, fluorescent enzyme substrates, including C12-FDG (5-dodecanoylaminofluorescein Di-β-D-Galactopyranoside, green) and DDAOG (9H-(1,3-dichloro-9,9-dimethylacridin-2-one-7-yl) β-D-Galactopyranoside, far-red) have previously appeared in the literature12,13,14,15. The chemical probe structure and optical characteristics of DDAOG are shown in Supplementary Figure S1. These cell-permeant probes permit the analysis of living (rather than fixed) cells, and fluorescent rather than chromogenic probes facilitate the use of rapid high-throughput fluorescent analysis platforms, including high-content screening instruments and flow cytometers. Sorting flow cytometers enable the recovery of enriched populations of living senescent cells from cell cultures or tumors for downstream analysis (e.g., western blotting, ELISA, or 'omics). Fluorescence analysis also provides a quantitative signal, allowing for more accurate determination of the percentage of senescent cells within a given sample. Additional fluorescent probes, including viability probes and fluorophore-labeled antibodies, can readily be added for multiplexed analysis of targets beyond SA-β-Gal.
Similar to DDAOG, C12-FDG is a fluorescent probe for SA-β-Gal, but its green fluorescent emission overlaps with intrinsic cellular AF, which arises during senescence due to the accumulation of lipofuscin aggregates in cells16. By utilizing the far-red DDAOG probe, green cellular AF can be used as a secondary parameter to confirm senescence17. This improves assay reliability by using a second marker in addition to SA-β-Gal, which can often be unreliable as a single marker for senescence18. As the detection of endogenous AF in senescent cells is a label-free approach, it is a rapid and simple way to expand the specificity of our DDAOG-based assay.
In this protocol, we demonstrate the use of DDAOG and AF as a rapid, dual-parameter flow cytometry assay for the identification of viable TIS tumor cells from in vitro cultures or isolated from drug-treated tumors established in mice (Figure 1). The protocol uses fluorophores compatible with a wide range of standard commercial flow cytometry analyzers and sorters (Table 1). Quantitation of the percentage of viable senescent cells using standard flow cytometry analysis is enabled. If desired, an optional immunolabeling step may be performed to evaluate cell surface antigens of interest concurrently with senescence. Identified senescent cells can also be enriched using standard fluorescence-activated cell sorting (FACS) methodology.
Figure 1: Experimental workflow. A schematic summarizing key points of the DDAOG assay. (A) A TIS-inducing drug is added to mammalian cultured cells or administered to tumor-bearing mice. Time is then allowed for the onset of TIS: for cells, 4 days following treatment; for mice, 22 days total, with three treatments every 5 days plus 7 days recovery. Cells are harvested or tumors are dissociated into suspension. (B) Samples are treated with Baf to adjust lysosomal pH for detection of SA-β-Gal for 30 min; then, DDAOG probe is added for 60 min to detect SA-β-Gal. Samples are washed 2x in PBS, and a viability stain is briefly added (15 min). Optionally, samples can be stained with fluorescent antibodies in open fluorescence channels and/or fixed for later analysis. (C) Samples are analyzed using a standard flow cytometer. Viable cells are visualized in dot plots showing red DDAOG (indicating SA-β-Gal) versus green autofluorescence (lipofuscin). A gate to determine the percentage of TIS cells is established based on untreated control samples (not shown). If a sorting cytometer (FACS) is used, TIS cells can be collected and placed back into culture for further in vitro assays or lysed and processed for molecular biology assays. Abbreviations: DDAO = 9H-(1,3-dichloro-9,9-dimethylacridin-2-one); DDAOG = DDAO-Galactoside; TIS = therapy-induced senescence; FL-Ab = fluorophore-conjugated antibody; Baf = Bafilomycin A1; SA-β-Gal = senescence-associated beta-galactosidase; PBS = phosphate-buffered saline; FACS = fluorescence-activated cell sorting. Please click here to view a larger version of this figure.
Fluorophore | Detects | Ex/Em (nm) | Cytometer laser (nm) | Cytometer detector / bandpass filter (nm) |
DDAOG | SA-β-Gal | 645/6601 | 640 | 670 / 30 |
AF | Lipofuscin | < 600 | 488 | 525 / 50 |
CV450 | Viability | 408/450 | 405 | 450 / 50 |
PE | Antibody/surface marker | 565/578 | 561 | 582 / 15 |
Table 1: Fluorophores and cytometer optical specifications. Cytometer specifications used in this protocol are listed for an instrument with a total of 4 lasers and 15 emission detectors. DDAOG detected at 645/660 nm is the form of the probe cleaved by SA-β-Gal1. Uncleaved DDAOG can exhibit low level fluorescence at 460/610 nm but is removed by wash steps in the protocol. Abbreviations: DDAO = 9H-(1,3-dichloro-9,9-dimethylacridin-2-one); DDAOG = DDAO-Galactoside; AF = autofluorescence; PE = phycoerythrin; SA-β-Gal = senescence-associated beta-galactosidase.
All animal work described was approved by the Institutional Animal Care and Use Committee at the University of Chicago.
1. Preparation and storage of stock solutions
NOTE: If cells will be flow-sorted, all solutions should be prepared using sterile techniques and filtered through a 0.22 µm filter device.
2. Induction of senescence by chemotherapy drugs in cultured cancer cells
NOTE: All cell manipulation steps in this section should be performed in a biosafety cabinet using sterile practices. This section is written for adherent cell types. Suspension cells may be used with appropriate modifications as noted.
3. Induction of senescence by chemotherapy drugs in tumors established in mice
NOTE: If tumor cells will be FACS-sorted, ensure sterility at each step by working in a biosafety cabinet and working with sterile instruments, procedures, and reagents.
4. DDAOG staining of SA-β-Gal in cell or tumor samples
5. (Optional) Immunostaining for cell surface markers in combination with DDAOG
NOTE: As with any flow cytometry experiment, single-stained control samples with DDAOG only and fluorescent antibody only should be prepared to determine crosstalk (if any) across fluorescence channels. If crosstalk is observed, standard flow cytometry compensation should be performed20.
6. Flow cytometer setup and data acquisition
7. Flow cytometry data analysis
NOTE: The workflow presented uses FlowJo software. Alternative flow cytometry data analysis software may be used if the key steps described in this section are similarly followed.
Several experiments were performed to demonstrate the comparability of DDAOG to X-Gal and C12-FDG for the detection of senescence by SA-β-Gal. First, X-Gal was used to stain senescent B16-F10 melanoma cells induced by ETO (Figure 2A). An intense blue color developed in a subset of ETO-treated cells, while other cells exhibited less intense blue staining. Morphology was enlarged in most ETO-treated cells. Staining ETO-treated cells with fluorescent SA-β-Gal substrate C12-FDG (green) or DDAOG (far-red) demonstrated comparable staining patterns and intensity variations to X-Gal (Figure 2B). However, green C12-FDG emission overlapped with cellular AF (Figure 2C), which is known to accumulate in senescent cells17. In contrast, AF was negligible in the far-red emission range of DDAOG.
Instead of using a fluorescence microscope to score and count senescent cells, it was more expedient to take advantage of the high-throughput capabilities of a flow cytometer to acquire data for thousands of cells per sample in a short time (<5 min per sample). First, a series of standard flow cytometry setup parameters were implemented to ensure optimal data acquisition (Figure 3). Following a typical approach, light scatter parameters were set to visualize the volume (forward scatter, FSC) and granularity (side scatter, SSC) of cells (Figure 3A). Here, we noted the trend of increased volume of ETO-treated cells, which agreed with the enlarged morphology typically observed for senescent cells using microscopy. A reduction in default area scaling settings (to 0.33-0.50 units depending on the cell type and treatment) was necessary to visualize more of the larger cells on the plot. For some cell lines/treatments, increased granularity (SSC) was also evident (data not shown). Overall, scatter evaluation was used as a quality control step to verify that cell scatter data appeared as expected, excessive cell debris was not present, and cells were being processed through the cytometer at appropriate flow rates (~100-1000 cells/s). As a quality control step used only for instrument setup, no gating or analysis was performed here.
A second (optional) step in the flow cytometry setup was to briefly analyze a sample of commercially available "rainbow" calibration particles to ensure the fluorescent detection voltages were set in acceptable ranges (Figure 3B). The brightest peak was set between 1 × 104 units and 1 × 105 units of relative fluorescent intensity in each channel, with clearly defined lower intensity peaks, sufficient separation between each peak, and no overlap of neighboring peaks. A control sample of 10,000 microspheres was recorded using these voltage settings. The microspheres were then used in this manner in each cytometry session to improve the uniformity of data acquisition over the course of the protocol.
Next, samples of vehicle-only or ETO-treated cells were visualized in each fluorescent channel, and gates were set. Cell viability gates were set based on the signal of CV450 viability stain in the violet channel (Figure 3C). Vehicle-only treated cells exhibited 88% viability, and ETO-treated cells exhibited 75% viability (in the final cell sample; additional dead cells were likely initially removed in discarded culture media and mechanically disintegrated during the staining process). Next, viable gated cells were visualized in the green (Figure 3D) and far-red (Figure 3E) emission channels. Gates for green AFHI and far-red DDAOGHI were set at <5% of vehicle-only cells, and these gates were then applied to ETO-treated cells. Using this approach, it was determined that 46% of ETO-treated cells were AFHI and 33% were DDAOGHI; these values were in the expected range based on literature and from results of numerous replicate experiments in our laboratory. Once the cytometer setup was complete, all cell samples in the experiment were run using identical data acquisition settings. Data for 10,000 events per sample were obtained.
Representative assay data is shown in Figure 4. B16-F10 murine melanoma cells or A549 human lung adenocarcinoma cells were used as the cancer cell line models. Each cell line was treated with a chemotherapy agent known to induce senescence (ETO or BLM) for 4 days to induce senescence or vehicle-only. Further, the known senolytic agent ABT-26321 was added to induced senescent cells for 2 days to demonstrate the specificity of the DDAOG probe. ABT-263 only samples were prepared as additional controls. Here, data are visualized as 2D dot plots with DDAOG (670 nm emission) versus AF (525 nm). The workflow from Figure 3 was used for cytometer setup, and a TIS gate was set such that <5% of vehicle-only cells scored as TIS. Results using B16-F10 cells (Figure 4A) showed that ETO induced TIS in 35% of viable B16-F10 cells (similar to the comparable data shown in Figure 2D,E) and the senolytic agent almost completely eliminated TIS cells (<2%). In A549 cells (Figure 4B), BLM induced TIS in 66% of viable cells, and ABT-263 reduced the percentage to 15%. ABT-263 alone was not toxic to untreated, proliferating cells.
We further aimed to demonstrate that fluorescent antibody co-staining was compatible with the DDAOG senescence assay to facilitate the screening of TIS-associated or novel surface markers. Here, B16-F10 mouse melanoma cells were again treated with TIS-inducing ETO (or vehicle) for 4 days. Then, cells were both stained with DDAOG to evaluate TIS and with a fluorescent antibody to detect the TIS-associated surface marker DPP422 (Figure 5). Anti-DPP4 conjugated to R-phycoerythrin (PE) was used, and we confirmed negligible overlap of PE with DDAOG and AF on the flow cytometer used (Supplementary Figure S2). A histogram of PE channel data (Figure 5A) for >5,000 viable cells showed that 42% of ETO-treated cells were DPP4+ (using the vehicle-only sample to set the positivity gate). Visualizing 2D dot plots for the same samples (Figure 5B,C) indicated that 44% of ETO-treated cells were double-positive for DDAOG (i.e., senescent) and DPP4 versus 4% of vehicle-only cells. These data demonstrate that live-cell staining with surface marker antibodies is possible in combination with the DDAOG senescence assay.
Potential concerns with any live cell staining method include cell death occurring during lengthy analysis sessions (>1 h) and how to most efficiently stain and analyze samples across many different time points (up to days apart). Solvent-based fixation of stained cells addresses both of these concerns, as samples can be fixed as soon as they are stained and then stored in the refrigerator until analysis in one temperature-stable batch. Thus, we tested whether live cells stained with DDAOG could be fixed with 100% methanol or 4% paraformaldehyde (PFA) and stored for up to 1 week at 4 °C. Undesirably, methanol fixation decreased the DDAOG signal and significantly reduced AF (data not shown); hence, the use of methanol as a fixation solvent should be avoided. However, fixation with PFA was much more successful, as seen in Figure 6 for cells fixed in 4% PFA for 10 min after staining with DDAOG. Compared to unfixed control samples (Figure 6A; untreated, 5% and BLM, 67% DDAOGHI AFHI), fixed samples (Figure 6B) exhibited slightly higher background in untreated cells (9%) and also a higher percentage of cells scoring as senescent in BLM-treated cells (80%). This effect was also seen in fixed samples stored overnight (Figure 6C; untreated, 12% and BLM, 72%) and stored for 1 week at 4 °C (Figure 6D; 14% and 70%). Despite the slight increases in fluorescence caused by PFA fixation, the induction of senescence by BLM was still evident in all fixed samples versus the matched untreated sample. Further, cells remained intact in storage with only slight deterioration by Day 7, and no problematic aggregation was observed. We conclude that the convenience of being able to fix and store cell samples for later batched analysis will justify tolerating the slightly higher background caused by PFA fixation, particularly in experiments with many samples or time points.
A common challenge in cell-based senescence research is the heterogeneous onset of senescence in cell populations. Here, we show that DDAOG can be used for FACS of viable senescent cells and that collected cells survive in culture for downstream in vitro assays (Figure 7). To sort cells by FACS, cells are treated and stained, as described, and sorted by a FACS-capable flow cytometer using the DDAOG versus AF gating strategy shown here (Figure 7A). As a viability probe is not used here due to long-term toxicity concerns, we recommend stringent scatter gating to be performed prior to the final senescent cell gating (Supplementary Figure S3) to eliminate false-positive debris from the collected cells. These are standard FACS gating procedures that should be familiar to a moderately experienced user and can be rapidly established using software at the instrument (<10 min).
After sorting a sample of BLM-treated A549 cells, we returned the cells to culture in standard multiwell dishes for 5 days (n = 6 replicates) at 10 × 103 cells/cm2. Unsorted controls were plated at the same density, grown alongside, and untreated and BLM-treated cells were included. Cells were observed daily; for sorted cells, no significant cell death or return to proliferation was observed. Sorted cells remained sparse (i.e., not proliferating) over the course of 5 days in culture, while untreated and BLM-treated cells became confluent as shown. On Day 5, cells were fixed in PFA and stained for morphology and proliferation markers (Figure 7B). The characteristic enlarged morphology of sorted cells was revealed by fluorescent Phalloidin staining of filamentous actin, with DAPI staining to show enlarged nuclei. Sorted cells were very large in diameter (>10 µm) with a characteristic rounded appearance. As expected, Ki67 antibody staining revealed a complete loss of the proliferation marker in the sorted sample versus a partial loss in the unsorted BLM-treated sample, and normal levels of Ki67 were seen in many cells in the untreated-unsorted sample. At least three images were taken per well using uniform imaging settings across samples. Representative images are shown (Figure 7B).
Finally, we assessed whether senescent cells arising in tumors established in mice treated with chemotherapeutic drugs could be identified using DDAOG. B16-F10 melanoma tumors were induced in the flank of C57/BL6 mice, which were then treated three times (i.p., every 5 days) with saline only (Figure 8A), DOX (Figure 8B), or PLD (Figure 8C). After the third treatment, the mice recovered for 7 days to allow the onset of senescence and were then sacrificed and tumors excised. Tumors were halved, with half used to prepare frozen tissue slides for X-Gal staining (Figure 8) and half dissociated into single-cell suspensions and stained with DDAOG (Figure 8 and Supplementary Figure S4). X-Gal staining in tissues was rather weak despite a long staining duration (72 h), but blue staining was evident in DOX and PLD tumors upon close inspection, particularly in tumors that also scored positive for senescence by DDAOG flow assay (Figure 8B,C). At least three images were taken per tissue slide and representative images are shown.
Compared to X-Gal, DDAOG was a more sensitive and accurate way to quantify senescence in tumors (Figure 8). For DDAOG tumor analysis, the saline-treated tumor with the highest fluorescence background was used to set the senescence gate at <5%, such that the other saline-treated tumors did not score above 5% senescence. This senescence gate was then batch-applied to gated viable cells from all tumors. Both chemotherapy agents induced senescence in some tumors (two of five tumors for DOX, three of five for PLD), with percentages of senescent tumor cells varying from 3% to 36% per tumor. Cytometry data plots for all 15 tumors are shown in Supplementary Figure S4, and a summary of tumor cytometry data is shown in Figure 8D (n = 5; (*) p < 0.05 vs. saline-only control group by the F test of variance). We conclude that DDAOG versus AF flow cytometry is an acceptable method for screening tumors and chemotherapy agents for the induction of senescence in vivo.
Figure 2: DDAOG is a sensitive and specific stain for SA-β-Gal. (A) Conventional staining for SA-β-Gal using X-Gal showing untreated (left) or ETO-treated (right) B16-F10 murine melanoma cells. Senescence induced by ETO: cells exhibit enlarged morphology and blue staining due to cleavage of X-Gal by elevated SA-β-Gal. Scale bars = 10 µm. (B) Fluorescent staining for SA-β-Gal in ETO-treated cells using either C12-FDG (515 nm emission, green) or DDAOG (660 nm, red). Staining distribution is similar for both probes, indicating that DDAOG detects SA-β-Gal in a similar manner to C12-FDG. (C) Evaluation of AF in unstained, ETO-treated cells in either the green (525 nm emission, left) or far-red (660 nm, right) emission channel. AF from lipofuscin is high in the green emission channel and negligible in the far-red channel. Exposure time = 2,000 ms for each image. Scale bars = 10 µm. This figure is reprinted from Flor and Kron23. Abbreviations: DDAO = 9H-(1,3-dichloro-9,9-dimethylacridin-2-one); DDAOG = DDAO-Galactoside; X-Gal = 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside; UNT = untreated; ETO = etoposide; AF = autofluorescence; C12-FDG = 5-dodecanoylaminofluorescein di-β-D-galactopyranoside; SA-β-Gal = senescence-associated beta-galactosidase. Please click here to view a larger version of this figure.
Figure 3: Flow cytometer data acquisition setup. (A) Representative scatter plot distribution of cells. FSC-A is a readout of cell volume, and SSC-A indicates cellular granularity. Left panel, vehicle-only treatment of A549 cells. Right panel, ETO treatment to induce senescence. Note the trend toward enlarged cell volume of ETO-treated cells, in agreement with the enlarged morphology evident from microscopy. (B) Optional use of 5-peak commercial "rainbow" fluorescent calibration microspheres to set detector voltages of the cytometer. In each fluorescence channel used, the maximum peak should be set ≤1 x 105 relative fluorescent units by adjusting the cytometer detector voltage while samples are running. Left panel, BV421 (violet) channel; center, FITC (green) channel; right, APC (red) channel. The five fluorescent peaks should exhibit distinct spacing as shown. (C–E) Representative single-channel fluorescence data for (C) viability staining using violet CV450 dye; (D) green autofluorescence of cells; (E) far-red signal from DDAOG, detecting SA-β-Gal. Darker color histogram, vehicle-only treatment; lighter color histogram, ETO treatment. Parent gates are indicated above each plot. Abbreviations: FSC-A = forward scatter-peak area; SSC-A = side scatter-peak area; ETO = etoposide; BV421-A = Brilliant Violet 421 channel peak area; FITC-A = fluorescein isothiocyanate channel peak area; APC-A = allophycocyanin channel peak area; AF = autofluorescence; DDAO = 9H-(1,3-dichloro-9,9-dimethylacridin-2-one); DDAOG = DDAO-Galactoside; VEH = vehicle. Please click here to view a larger version of this figure.
Figure 4: Representative data for flow cytometry senescence assay. (A) B16-F10 murine melanoma cells treated with (upper left) vehicle only, (upper right) ETO, (lower left) ABT-263 (1 µM) only, or (lower right) ETO plus ABT-263. (B) A549 human lung adenocarcinoma cells treated with (upper left) vehicle only, (upper right) BLM, (lower left) ABT-263 only, or (lower right) BLM plus ABT-263. (A,B) Rectangular gates in upper right quadrants of all plots define senescent cells (DDAOGHI AFHI). The percentage of senescent cells (of total viable cells per sample) is indicated on each plot. Abbreviations: DDAO = 9H-(1,3-dichloro-9,9-dimethylacridin-2-one); DDAOG = DDAO-Galactoside; ETO = etoposide; BLM = bleomycin; VEH = vehicle; TIS = therapy-induced senescence. Please click here to view a larger version of this figure.
Figure 5: Co-staining of an example cell surface marker with senescence assay. (A) Immunodetection of senescence marker DPP4 on the surface of viable B16-F10 cultured cells; dark orange, vehicle only; light orange, ETO. (B,C) Center and right panels: cells co-stained with DDAOG and anti-DPP4:PE. DDAOGHI DPP4HI cells are contained within rectangular gates shown on the plots, and the percentage of double-positive cells per sample is indicated. Shown: viable-gated cells. Abbreviations: DDAO = 9H-(1,3-dichloro-9,9-dimethylacridin-2-one); DDAOG = DDAO-Galactoside; ETO = etoposide; VEH = vehicle; DPP4 = dipeptidyl peptidase 4. Please click here to view a larger version of this figure.
Figure 6: Poststaining fixation and storage of DDAOG-stained cell samples for later analysis. (A) Control, unfixed samples of live A549 cells either untreated (left) or treated with BLM (right) to induce senescence, and then stained and immediately analyzed using the DDAOG protocol (without fixation, Day 0). (B) Samples prepared as in (A) and then immediately fixed in 4% paraformaldehyde and analyzed (Day 0). (C) Samples prepared as in (A), immediately fixed in 4% paraformaldehyde, and stored overnight at 4 °C prior to analysis. (D) Samples prepared as in (A), immediately fixed, and stored for 7 days prior to analysis. Rectangular gates in the upper right quadrants of all plots define senescent cells (DDAOGHI AFHI). The percentage of senescent cells is indicated on each plot. Abbreviations: TIS = therapy-induced senescence; DDAO = 9H-(1,3-dichloro-9,9-dimethylacridin-2-one); DDAOG = DDAO-Galactoside; BLM = bleomycin; AF = autofluorescence. Please click here to view a larger version of this figure.
Figure 7: Flow cytometric sorting and validation of enriched senescent cell populations. (A) Flow cytometric sorting data showing (left) the untreated, DDAOG-stained control used to set the gate for senescent cell sorting (<5% senescent cells in gated region) and (right) the BLM-treated, DDAOG-stained sample that was sorted using the sort gate as shown. (B) Fluorescence microscopy images for (left column) cells stained with Phalloidin-Alexa Fluor 647 (orange pseudocolor) to detect F-actin and DAPI (blue) to counterstain nuclei, demonstrating enlarged morphology of senescent cells, or (right column) cells stained with rabbit Ki67 antibody and anti-rabbit Alexa Fluor 594 to detect loss of proliferation in senescent cells. Top row, untreated and unsorted cells. Center row, BLM-treated and unsorted cells. Lower row, BLM-treated and sorted cells. Scale bars = 10 µm. Abbreviations: TIS = therapy-induced senescence; DDAO = 9H-(1,3-dichloro-9,9-dimethylacridin-2-one); DDAOG = DDAO-Galactoside; BLM = bleomycin; UNT = untreated; DAPI = 4',6-diamidino-2-phenylindole. Please click here to view a larger version of this figure.
Figure 8: Quantification of senescence in tumors from mice treated with chemotherapy drugs. B16-F10 melanoma tumors were established in the flanks of C67BL/6 mice, which were then treated with three doses of (A) saline, (B) DOX, or (C) PLD every 5 days plus 1 week to allow the onset of senescence. Tumors were excised and halved; frozen tissue slides were prepared from one half for X-Gal staining (top row), and the other half was dissociated for DDAOG staining (lower row). In X-Gal-stained images, blue cells are SA-β-Gal HI senescent cells, while brown regions are due to melanin in tumors. Representative results are shown for DOX and PLD from two tumors per group that exhibited senescence. All saline-only tumors exhibited negligible senescence. (D) Quantification of senescence in drug-treated tumors from mice. (*) p < 0.05 by F test of variance, n = 5 mice per group. Error bars, mean ± SEM. Abbreviations: DDAO = 9H-(1,3-dichloro-9,9-dimethylacridin-2-one); DDAOG = DDAO-Galactoside; DOX = doxorubicin; PLD = PEGylated liposomal doxorubicin; X-Gal = 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside; AF = autofluorescence. Please click here to view a larger version of this figure.
Supplementary Figure S1: Chemical structure of the DDAOG probe. DDAOG is a conjugate of 7-hydroxy-9H-(1,3-dichloro-9,9-dimethylacridin-2-one) and beta galactoside. When cleaved by beta-galactosidase, the hydrolyzed cleavage product exhibits a 50 nm far-red fluorescence emssion shift, enabling its specific detection with excitation above 600 nm. Abbreviations: DDAO = 9H-(1,3-dichloro-9,9-dimethylacridin-2-one); DDAOG = DDAO-Galactoside. Please click here to download this File.
Supplementary Figure S2: Spectral scan of DDAOG crosstalk with other fluorescent channels of the flow cytometer. To identify available channels for the detection of fluorescent antibodies, a spectral scan was performed on a 4-laser, 15-channel flow cytometer using A549 cells treated with BLM and stained with DDAOG (red) or unstained (black). Data in every channel of the flow cytometer were acquired for 10,000 cells. (A) Emission channels of the 405 nm laser: from left to right, BV421, BV510, BV605, BV660, and BV711. The crosstalk observed in BV605, BV660, and BV711 channels makes them unsuitable for co-staining with DDAOG without compensation. BV421 and BV510 are suitable for co-staining (note that BV421 is typically used for viability staining). (B) Emission channels of the 488 nm laser: FITC and PerCP-Cy5. High crosstalk was observed for the PerCP-Cy5 channel. FITC is suitable for co-staining; however, note that the FITC channel is typically used in the DDAOG assay for the evaluation of green emission AF. (C) Emission channels of the 561 nm laser: PE, PE-Dazzle 594, PE-Cy5, PE-Cy5.5, and PE-Cy7. The PE and PE-Dazzle 594 channels are suitable for the detection of antibodies labeled with these fluorophores (PE is demonstrated for the detection of DPP4 in this study). (D) Emission channels of the 640 nm laser: APC, APC-H700, and APC-Cy7. The DDAOG signal of the senescent cells is visible in the APC channel (47.8% senescent). The signal overlaps into the APC-H700 and APC-Cy7 channels, making them unsuitable for co-staining without significant spectral compensation. Abbreviations: BLM = bleomycin; BV = Brilliant Violet; FITC = fluorescein isothiocyanate; PerCP = peridinin-chlorophyll protein; PE = phycoerythrin; DZ594 = Dazzle 594; APC = allophyocyanin; AF = autofluorescence; DDAO = 9H-(1,3-dichloro-9,9-dimethylacridin-2-one); DDAOG = DDAO-Galactoside. Please click here to download this File.
Supplementary Figure S3: Cell gating strategy for FACS sorting of senescent cells. Passing a suspension of cells through a sensitive FACS cytometer can require additional gating to ensure the purity of sorting and optimal instrument functionality. An example strategy is shown here. Other strategies are possible depending on manufacturer recommendations for the FACS cytometer being used. Left column, vehicle-only treated cell control. Right column, A549 cells treated with BLM to induce senescence to be sorted. (A) FSC-A (cell volume) versus SSC-A (cell granularity) gating of intact cells. The intact gate eliminates cell debris from the sorted sample. (B) FSC-A versus FSC-H purity gating; removes doublets and debris. (C) SSC-A versus SSC-H purity gating; removes doublets and debris. (D) Gating for the population of interest using the APC-A channel (637 nm excitation, 670 nm ± 30 nm emission, used for DDAOG) versus FITC-A channel (488 nm excitation, 530 nm ± 30 nm emission, used for AF); removes possible staining artifacts. (E) Senescent cell gating for final sorting. Abbreviations: FACS = fluorescence-activated cell sorting; BLM = bleomycin; FSC-A = forward scatter-peak area; SSC-A = side scatter-peak area; FSC-H = forward scatter-peak height; SSC-H = side scatter-peak height. Please click here to download this File.
Supplementary Figure S4: DDAOG senescence flow cytometry staining of tumors. Flow cytometry data for ≥50,000 viable tumor cells. Senescent cell gate, upper right quadrant of each plot. The percentage of senescent (of viable) tumor cells is shown within the gate. Tumor treatments included (A) saline-only; (B) DOX; and (C) PLD. Mice were treated three times, once every 5 days, with 7 days for recovery to allow the onset of senescence before sacrifice. Five tumors per condition were analyzed. Abbreviations: AF = autofluorescence; DDAO = 9H-(1,3-dichloro-9,9-dimethylacridin-2-one); DDAOG = DDAO-Galactoside; DOX = doxorubicin; PLD = PEGylated liposomal doxorubicin. Please click here to download this File.
Over the last decade or so, flow cytometry has become a more common assay platform in cancer research due to the emerging popularity of tumor immunology, the development of lower-cost flow cytometers, and the improvement of shared instrumentation facilities at academic institutions. Multicolor assays are now standard, as most newer instruments are equipped with violet, blue-green, and red to far-red optical arrays. Thus, this DDAOG protocol is likely to be compatible with a wide array of flow cytometers. Of course, any flow cytometer should be user-evaluated. Particular care should be taken when adding additional fluorophores (e.g., fluorescent, conjugated antibodies) to the DDAOG assay. An evaluation of fluorophore crosstalk between channels should be conducted using single-stained controls visualized in all other relevant channels. If overlap is observed, spectral compensation can be performed20 for correction following typical methods.
The findings shown herein are primarily intended to demonstrate that the DDAOG flow cytometry assay can produce rapid, quantitative, easy-to-interpret results for TIS induced by chemotherapy drugs in cells or tumors. The agents used here, including ETO23,24, DOX7,25, and BLM26,27, have been documented to induce TIS in various cancer cell lines24. To demonstrate the specificity of the DDAOG probe, the known senolytic agent ABT-26321 was demonstrated to selectively eliminate senescent cells in culture. This paper demonstrates the use of one murine (B16-F10) and one human (A549) cultured cancer cell line, as well as B16-F10 tumors established in mice. However, any cells that express β-Gal and retain viability through a standard flow cytometry sample preparation can be used. Certain cell types may be more fragile or less prone to TIS, and this should be evaluated before embarking upon a large screen or study. If cells disintegrate in preparation, viability is poor, or TIS is much lower than expected using positive agent controls, the cell type may not be an ideal model for studies of senescence. The agents and cell lines shown here can be used by other groups as controls in further screening of potential TIS-inducing agents or novel senolytics, which remains an active goal in the field.
Toxicity induced by chemotherapy drugs can vary across cell types and affect assay results. If the primary observed effect of the agent and/or concentration used is acute cell death, overall senescence may be minimal. In vivo, high tumor necrosis or systemic toxicity in animals is to be avoided by lowering the agent dosage. The key to assay success is the testing of a range of agent concentrations, with careful inspection of viability assay data (e.g., as provided by CV450 within the DDAOG assay). In vitro, if many dead cells detach from the plate during treatment, including the culture medium in the analyzed sample, it is important to evaluate cell death in total. CV450 is not the sole viability stain that is compatible with this assay; other violet/blue emitting fluorophores may be used. Fixable viability probes can also be used if the user plans to fix the stained samples with PFA, provided the probe fluorescence does not significantly overlap with DDAOG or AF (or the user conducts spectral compensation to correct for overlap). In some cases where agent toxicity is low and cells are robust, gating intact cells by light scatter (FSC vs. SSC) can be sufficient to isolate "viable" cells for analysis.
A key step in this in vitro protocol is to plate cells in the lower range of log growth density (typically 2 × 103-5 × 103 cells/cm2) to allow rapid initial proliferation, which facilitates the uptake of the chemotherapy drug by most cells. Once treated, it is also important to allow time for the onset of TIS: in vitro, 4 days ± 1 day in the presence of the drug; in vivo, 7 days of recovery following the final chemotherapy treatment. After staining with DDAOG as described, quantitative analysis of cytometric data should be performed as shown, i.e., gating DDAOG versus AF cells in each sample to determine the percentage of senescent cells (of total viable). Optional steps include the use of fluorescent "rainbow" calibration microspheres to standardize flow cytometry, PFA fixation of stained samples, co-immunostaining for surface markers, and flow sorting to enrich senescent cells for downstream assays. However, each of these optional steps provides key advantages in certain applications. The calibration microspheres standardize the cytometry setup across multiple sessions and allow the user to initially set voltages in a useful range for senescence and viability detection, with minimal adjustments thereafter. PFA fixation of samples stabilizes the cells and allows batched analysis of large sets of cells at a later time. Co-immunostaining for surface markers can be used to screen for novel senescence-related proteins and immune interactions. In future studies, we plan to validate the co-staining of intracellular senescence markers with DDAOG.
Sorting TIS cells by FACS allows for the enrichment of viable TIS cells from heterogeneous populations, which can confound readouts for downstream assays such as western blotting, proteomics, or transcriptomics. After sorting, no significant toxicity of DDAOG was observed in TIS cells placed back into culture for up to 5 days. However, it should be taken into consideration that sorted cells have been treated with Baf, internalized DDAOG (which is cleaved by SA-β-Gal to DDAO, an acridine dye), and been subjected to mechanical stress passing through the FACS instrument. Therefore, certain biological alterations not related to senescence may be present in the sorted cells. However, in this study, the sorted cells retained strong features of senescence and provided characteristic proteomics and transcriptomics results28 when compared to unsorted controls. Despite the somewhat obvious utility of using FACS to collect and analyze senescent cells from tumors, this procedure has rarely been used in the literature. Some groups have used p16Ink4a luciferase or fluorescent reporters to identify senescent cells in mouse tissues, with fluorescent reporters enabling FACS sorting in some studies29,30. Findings generally agree that, regardless of the induction agent or tumor type, TIS in tumors is a partial to rare occurrence, rarely reaching 100% of tumor cells31. FACS sorting of cells using the DDAOG method readily allows the collection of rare TIS cells from tumors, without the need to express transgenic constructs.
Currently, most senescence research in mouse models is conducted ex vivo using a combination of X-Gal and immunohistochemistry markers32,33. Yet, assessing TIS ex vivo using tumor tissues can be a lengthy process, particularly when X-Gal is used. This procedure requires tumor cryopreservation, cryosectioning onto slides, X-Gal staining, cover glass mounting, drying, imaging, and scoring "blue" cells-a multi-day approach at minimum. Immunohistochemistry is not much faster or easier, and different tissue sections must be used and scored for each marker plus X-Gal in each tumor, unless the multiplexed analysis is optimized, adding more time at the front-end of the process. Tissue sections sample only one thin cross-section of the tumor, while senescent cells (like many cell types) may be distributed unevenly in 3D space throughout tumors29. It is urgently needed in the field to move away from slow, outdated histology methods toward a more rapid and readily quantifiable senescence assay for tumors. Using DDAOG flow cytometry, a set of 15 tumors could be dissociated and stained to obtain conclusive, quantitative senescence data in less than 1 day of hands-on time following tumor harvest. One half of the tumor was processed for each sample, improving the sampling of 3D space per tumor. This DDAOG flow cytometry approach is significantly faster and more reliable than tissue slide-based methods for the evaluation of TIS in tumors.
With its many advantages over X-Gal and other methods, we advocate for the DDAOG protocol to become the new gold standard senescence assay. It uses both the conventionally accepted senescence marker, SA-β-Gal, and label-free detection of age-associated AF as a second marker. These fluorophores are compatible with most standard flow cytometers. The assay includes a viability stain to exclude dead cells and debris using samples readily obtained from drug-treated cultured cell lines or tumors. Live cell samples can optionally be solvent-fixed or cryopreserved to facilitate the batched analysis of large studies, e.g., using time points to evaluate the onset of TIS over time using multiple agents The staining process takes less than a half-day of lab work to complete, and data acquisition is typically <5 min per sample. Data analysis is similarly rapid and straightforward, producing quantitative data on the percentage of TIS cells per sample without tedious cell scoring and counting. Samples can be FACS-sorted to recover enriched populations of TIS cells, which reduces noise from cellular heterogeneity in downstream analyses. We believe the DDAOG assay can, in many cases, replace X-Gal, facilitating the screening of TIS-inducing agents and senolytics in vitro and in vivo, leading to faster and more reliable discoveries in the field of senescence research.
The authors have nothing to disclose.
We thank the Cytometry and Antibody Core Facility at the University of Chicago for support on flow cytometry instrumentation. The Animal Research Center at the University of Chicago provided animal housing.
Bafilomycin A1 | Research Products International | B40500 | |
Bleomycin sulfate | Cayman | 13877 | |
Bovine serum albumin (BSA) | US Biological | A1380 | |
Calcein Violet 450 AM viability dye | ThermoFisher Scientific | 65-0854-39 | eBioscience |
DPP4 antibody, PE conjugate | Biolegend | 137803 | Clone H194-112 |
Cell line: A549 human lung adenocarcinoma | American Type Culture Collection | CCL-185 | |
Cell line: B16-F10 mouse melanoma | American Type Culture Collection | CRL-6475 | |
Cell scraper | Corning | 3008 | |
Cell strainers, 100 µm | Falcon | 352360 | |
DDAO-Galactoside | Life Technologies | D6488 | |
DMEM medium 1x | Life Technologies | 11960-069 | |
DMSO | Sigma | D2438 | |
DNAse I | Sigma | DN25 | |
Doxorubicin, hydrochloride injection (USP) | Pfizer | NDC 0069-3032-20 | |
Doxorubicin, PEGylated liposomal (USP) | Sun Pharmaceutical | NDC 47335-049-40 | |
EDTA 0.5 M | Life Technologies | 15575-038 | |
Etoposide | Cayman | 12092 | |
FBS | Omega | FB-11 | |
Fc receptor blocking reagent | Biolegend | 101320 | Anti-mouse CD16/32 |
Flow cytometer (cell analyzer) | Becton Dickinson (BD) | Various | LSRFortessa |
Flow cytometer (cell sorter) | Becton Dickinson (BD) | Various | FACSAria |
GlutaMax 100x | Life Technologies | 35050061 | |
HEPES 1 M | Lonza | BW17737 | |
Liberase TL | Sigma | 5401020001 | Roche |
Paraformaldehyde 16% | Electron Microscopy Sciences | 15710 | |
Penicillin/Streptomycin 100x | Life Technologies | 15140122 | |
Phosphate buffered saline (PBS) 1x | Corning | MT21031CV | Dulbecco's PBS (without calcium and magnesium) |
Rainbow calibration particles, ultra kit | SpheroTech | UCRP-38-2K | 3.5-3.9 µm, 2E6/mL |
RPMI-1640 medium 1x | Life Technologies | 11875-119 | |
Sodium chloride 0.9% (USP) | Baxter Healthcare Corporation | 2B1324 | |
Software for cytometer data acquisition, "FACSDiva" | Becton Dickinson (BD) | n/a | Contact BD for license |
Software for cytometer data analysis, "FlowJo" | TreeStar | n/a | Contact TreeStar for license |
Trypsin-EDTA 0.25% | Life Technologies | 25200-114 |